htslib 0.2.5 → 0.2.6

data/lib/hts/bam.rb

@@ -59,7 +59,7 @@ module HTS
       @start_position = tell
     end
 
-    def build_index(index_name = nil, min_shift: 0)
+    def build_index(index_name = nil, min_shift: 0, threads: 2)
       check_closed
 
       if index_name
@@ -67,10 +67,15 @@ module HTS
       else
         warn "Create index for #{@file_name}"
       end
-      r = LibHTS.sam_index_build3(@file_name, index_name, min_shift, @nthreads)
-      raise "Failed to build index for #{@file_name}" if r < 0
-
-      self
+      case LibHTS.sam_index_build3(@file_name, index_name, min_shift, (@nthreads || threads))
+      when 0 # successful
+      when -1 then raise "indexing failed"
+      when -2 then raise "opening #{@file_name} failed"
+      when -3 then raise "format not indexable"
+      when -4 then raise "failed to create and/or save the index"
+      else raise "unknown error"
+      end
+      self # for method chaining
     end
 
     def load_index(index_name = nil)
@@ -95,11 +100,6 @@ module HTS
       super
     end
 
-    def fai=(fai)
-      check_closed
-      LibHTS.hts_set_fai_filename(@hts_file, fai) > 0 || raise
-    end
-
     def write_header(header)
       check_closed
 
@@ -107,14 +107,22 @@ module HTS
       LibHTS.sam_hdr_write(@hts_file, header)
     end
 
-    def write(aln)
+    def header=(header)
+      write_header(header)
+    end
+
+    def write(record)
       check_closed
 
-      aln_dup = aln.dup
-      r = LibHTS.sam_write1(@hts_file, header, aln_dup)
+      # record = record.dup
+      r = LibHTS.sam_write1(@hts_file, header, record)
       raise "Failed to write record" if r < 0
     end
 
+    def <<(aln)
+      write(aln)
+    end
+
     def each(copy: false, &block)
       if copy
         each_record_copy(&block)

data/lib/hts/bcf/format.rb

@@ -8,30 +8,10 @@ module HTS
         @p1 = FFI::MemoryPointer.new(:pointer) # FIXME: naming
       end
 
-      # For compatibility with HTS.cr.
-      def get_int(key)
-        get(key, :int)
-      end
-
-      # For compatibility with HTS.cr.
-      def get_float(key)
-        get(key, :float)
-      end
-
-      # For compatibility with HTS.cr.
-      def get_flag(key)
-        get(key, :flag)
-      end
-
-      # For compatibility with HTS.cr.
-      def get_string(key)
-        get(key, :string)
-      end
-
-      def [](key)
-        get(key)
-      end
-
+      # @note: Why is this method named "get" instead of "fetch"?
+      # This is for compatibility with the Crystal language
+      # which provides methods like `get_int`, `get_float`, etc.
+      # I think they are better than `fetch_int`` and `fetch_float`.
       def get(key, type = nil)
         n = FFI::MemoryPointer.new(:int)
         p1 = @p1
@@ -73,6 +53,30 @@ module HTS
         end
       end
 
+      # For compatibility with HTS.cr.
+      def get_int(key)
+        get(key, :int)
+      end
+
+      # For compatibility with HTS.cr.
+      def get_float(key)
+        get(key, :float)
+      end
+
+      # For compatibility with HTS.cr.
+      def get_flag(key)
+        get(key, :flag)
+      end
+
+      # For compatibility with HTS.cr.
+      def get_string(key)
+        get(key, :string)
+      end
+
+      def [](key)
+        get(key)
+      end
+
       def fields
         ids.map do |id|
           name = LibHTS.bcf_hdr_int2id(@record.header.struct, LibHTS::BCF_DT_ID, id)

data/lib/hts/bcf/info.rb

@@ -9,31 +9,11 @@ module HTS
         @p1 = FFI::MemoryPointer.new(:pointer) # FIXME: naming
       end
 
-      # For compatibility with HTS.cr.
-      def get_int(key)
-        get(key, :int)
-      end
-
-      # For compatibility with HTS.cr.
-      def get_float(key)
-        get(key, :float)
-      end
-
-      # For compatibility with HTS.cr.
-      def get_string(key)
-        get(key, :string)
-      end
-
-      # For compatibility with HTS.cr.
-      def get_flag(key)
-        get(key, :flag)
-      end
-
-      def [](key)
-        get(key)
-      end
-
       # @note Specify the type. If you don't specify a type, it will still work, but it will be slower.
+      # @note: Why is this method named "get" instead of "fetch"?
+      # This is for compatibility with the Crystal language
+      # which provides methods like `get_int`, `get_float`, etc.
+      # I think they are better than `fetch_int`` and `fetch_float`.
       def get(key, type = nil)
         n = FFI::MemoryPointer.new(:int)
         p1 = @p1
@@ -70,6 +50,30 @@ module HTS
         end
       end
 
+      # For compatibility with HTS.cr.
+      def get_int(key)
+        get(key, :int)
+      end
+
+      # For compatibility with HTS.cr.
+      def get_float(key)
+        get(key, :float)
+      end
+
+      # For compatibility with HTS.cr.
+      def get_string(key)
+        get(key, :string)
+      end
+
+      # For compatibility with HTS.cr.
+      def get_flag(key)
+        get(key, :flag)
+      end
+
+      def [](key)
+        get(key)
+      end
+
       # FIXME: naming? room for improvement.
       def fields
         keys.map do |key|

data/lib/hts/bcf.rb

@@ -54,7 +54,7 @@ module HTS
       @start_position = tell
     end
 
-    def build_index(index_name = nil, min_shift: 14)
+    def build_index(index_name = nil, min_shift: 14, threads: 2)
       check_closed
 
       if index_name
@@ -62,10 +62,15 @@ module HTS
       else
         warn "Create index for #{@file_name}"
       end
-      r = LibHTS.bcf_index_build3(@file_name, index_name, min_shift, @nthreads)
-      raise "Failed to build index for #{@file_name}" if r < 0
-
-      self
+      case LibHTS.bcf_index_build3(@file_name, index_name, min_shift, (@nthreads || threads))
+      when 0 # sccessful
+      when -1 then raise "indexing failed"
+      when -2 then raise "opening #{@file_name} failed"
+      when -3 then raise "format not indexable"
+      when -4 then raise "failed to create and/or save the index"
+      else raise "unknown error"
+      end
+      self # for method chaining
     end
 
     def load_index(index_name = nil)
@@ -97,15 +102,21 @@ module HTS
       LibHTS.bcf_hdr_write(@hts_file, header)
     end
 
-    def write(var)
+    def header=(header)
+      write_header(header)
+    end
+
+    def write(record)
       check_closed
 
-      var_dup = var.dup
-      r = LibHTS.bcf_write(@hts_file, header, var_dup)
+      # record = record.dup
+      r = LibHTS.bcf_write(@hts_file, header, record)
       raise "Failed to write record" if r < 0
     end
 
-    # Close the current file.
+    def <<(var)
+      write(var)
+    end
 
     def nsamples
       check_closed
@@ -164,13 +175,13 @@ module HTS
     def info(key = nil)
       check_closed
       position = tell
-      if key
-        ary = map { |r| r.info(key) }
-      else
-        raise NotImplementedError
-        # ary = each_copy.map { |r| r.info }
-        # ary = map { |r| r.info.clone }
-      end
+      raise NotImplementedError unless key
+
+      ary = map { |r| r.info(key) }
+
+      # ary = each_copy.map { |r| r.info }
+      # ary = map { |r| r.info.clone }
+
       seek(position)
       ary
     end
@@ -178,13 +189,13 @@ module HTS
     def format(key = nil)
       check_closed
       position = tell
-      if key
-        ary = map { |r| r.format(key) }
-      else
-        raise NotImplementedError
-        # ary = each_copy.map { |r| r.format }
-        # ary = map { |r| r.format.clone }
-      end
+      raise NotImplementedError unless key
+
+      ary = map { |r| r.format(key) }
+
+      # ary = each_copy.map { |r| r.format }
+      # ary = map { |r| r.format.clone }
+
       seek(position)
       ary
     end

data/lib/hts/faidx/sequence.rb

@@ -0,0 +1,64 @@
+require_relative "../faidx"
+
+module HTS
+  class Faidx
+    class Sequence
+      attr_reader :name, :faidx
+
+      def initialize(faidx, name)
+        raise unless faidx.has_key?(name)
+
+        @faidx = faidx
+        @name = name
+      end
+
+      def length
+        faidx.seq_len(name)
+      end
+      alias size length
+
+      def seq(start = nil, stop = nil)
+        faidx.seq(name, start, stop)
+      end
+
+      def qual(start = nil, stop = nil)
+        faidx.qual(name, start, stop)
+      end
+
+      def [](arg)
+        case arg
+        when Integer
+          if arg >= 0
+            start = arg
+            stop = arg
+          else
+            start = length + arg
+            stop = length + arg
+          end
+        when Range
+          arg = Range.new(arg.begin, arg.end + length, arg.exclude_end?) if arg.end&.<(0)
+          arg = Range.new(arg.begin + length, arg.end, arg.exclude_end?) if arg.begin&.<(0)
+          if arg.begin.nil?
+            if arg.end.nil?
+              start = nil
+              stop = nil
+            else
+              start = 0
+              stop = arg.exclude_end? ? arg.end - 1 : arg.end
+            end
+          elsif arg.end.nil?
+            # always include the first base
+            start = arg.begin
+            stop = length - 1
+          else
+            start = arg.begin
+            stop = arg.exclude_end? ? arg.end - 1 : arg.end
+          end
+        else
+          raise ArgumentError
+        end
+        seq(start, stop)
+      end
+    end
+  end
+end

data/lib/hts/faidx.rb

@@ -1,6 +1,7 @@
 # frozen_string_literal: true
 
 require_relative "../htslib"
+require_relative "faidx/sequence"
 
 module HTS
   class Faidx
@@ -25,7 +26,11 @@ module HTS
       end
 
       @file_name = file_name
-      @fai = LibHTS.fai_load(@file_name)
+      @fai = if [".fq", ".fastq"].include? File.extname(@file_name)
+               LibHTS.fai_load_format(@file_name, 2)
+             else
+               LibHTS.fai_load(@file_name)
+             end
 
       raise Errno::ENOENT, "Failed to open #{@file_name}" if @fai.null?
     end
@@ -38,10 +43,9 @@ module HTS
       LibHTS.fai_destroy(@fai)
     end
 
-    # FIXME: This doesn't seem to work as expected
-    # def closed?
-    #   @fai.null?
-    # end
+    def file_format
+      @fai[:format]
+    end
 
     # the number of sequences in the index.
     def length
@@ -50,31 +54,48 @@ module HTS
     alias size length
 
     # return the length of the requested chromosome.
-    def chrom_size(chrom)
+    def names
+      Array.new(length) { |i| LibHTS.faidx_iseq(@fai, i) }
+    end
+
+    alias keys names
+
+    def has_key?(key)
+      raise ArgumentError, "Expect chrom to be String or Symbol" unless key.is_a?(String) || key.is_a?(Symbol)
+
+      key = key.to_s
+      case LibHTS.faidx_has_seq(@fai, key)
+      when 1 then true
+      when 0 then false
+      else raise
+      end
+    end
+
+    def [](name)
+      name = LibHTS.faidx_iseq(@fai, name) if name.is_a?(Integer)
+      Sequence.new(self, name)
+    end
+
+    # return the length of the requested chromosome.
+    def seq_len(chrom)
       raise ArgumentError, "Expect chrom to be String or Symbol" unless chrom.is_a?(String) || chrom.is_a?(Symbol)
 
       chrom = chrom.to_s
       result = LibHTS.faidx_seq_len(@fai, chrom)
       result == -1 ? nil : result
     end
-    alias chrom_length chrom_size
 
-    # return the length of the requested chromosome.
-    def chrom_names
-      Array.new(length) { |i| LibHTS.faidx_iseq(@fai, i) }
-    end
-
-    # @overload fetch(name)
+    # @overload seq(name)
     #   Fetch the sequence as a String.
     #   @param name [String] chr1:0-10
-    # @overload fetch(name, start, stop)
+    # @overload seq(name, start, stop)
     #   Fetch the sequence as a String.
     #   @param name [String] the name of the chromosome
     #   @param start [Integer] the start position of the sequence (0-based)
     #   @param stop [Integer] the end position of the sequence (0-based)
     #   @return [String] the sequence
 
-    def seq(name, start = nil, stop = nil)
+    def fetch_seq(name, start = nil, stop = nil)
       name = name.to_s
       rlen = FFI::MemoryPointer.new(:int)
 
@@ -84,6 +105,7 @@ module HTS
         start < 0    && raise(ArgumentError, "Expect start to be >= 0")
         stop  < 0    && raise(ArgumentError, "Expect stop to be >= 0")
         start > stop && raise(ArgumentError, "Expect start to be <= stop")
+        stop >= seq_len(name) && raise(ArgumentError, "Expect stop to be < seq_len")
 
         result = LibHTS.faidx_fetch_seq(@fai, name, start, stop, rlen)
       end
@@ -95,5 +117,32 @@ module HTS
 
       result
     end
+
+    alias seq fetch_seq
+
+    def fetch_qual(name, start = nil, stop = nil)
+      name = name.to_s
+      rlen = FFI::MemoryPointer.new(:int)
+
+      if start.nil? && stop.nil?
+        result = LibHTS.fai_fetchqual(@fai, name, rlen)
+      else
+        start < 0    && raise(ArgumentError, "Expect start to be >= 0")
+        stop  < 0    && raise(ArgumentError, "Expect stop to be >= 0")
+        start > stop && raise(ArgumentError, "Expect start to be <= stop")
+        stop >= seq_len(name) && raise(ArgumentError, "Expect stop to be < seq_len")
+
+        result = LibHTS.faidx_fetch_qual(@fai, name, start, stop, rlen)
+      end
+
+      case rlen.read_int
+      when -2 then raise "Invalid chromosome name: #{name}"
+      when -1 then raise "Error fetching sequence: #{name}:#{start}-#{stop}"
+      end
+
+      result
+    end
+
+    alias qual fetch_qual
   end
 end

data/lib/hts/hts.rb

@@ -69,6 +69,11 @@ module HTS
       @hts_file.nil? || @hts_file.null?
     end
 
+    def fai=(fai)
+      check_closed
+      LibHTS.hts_set_fai_filename(@hts_file, fai) > 0 || raise
+    end
+
     def set_threads(n = nil)
       if n.nil?
         require "etc"
@@ -107,14 +112,12 @@ module HTS
     end
 
     def rewind
-      if @start_position
-        r = seek(@start_position)
-        raise "Failed to rewind: #{r}" if r < 0
+      raise "Cannot rewind: no start position" unless @start_position
 
-        tell
-      else
-        raise "Cannot rewind: no start position"
-      end
+      r = seek(@start_position)
+      raise "Failed to rewind: #{r}" if r < 0
+
+      tell
     end
 
     private

data/lib/hts/libhts/constants.rb

@@ -158,7 +158,7 @@ module HTS
         :specific,          :pointer
     end
 
-    class HtsIdx < FFI::Struct
+    class HtsIdx < FFI::Struct # FIXME: ManagedStruct
       layout \
         :fmt,            :int,
         :min_shift,      :int,
@@ -205,7 +205,7 @@ module HTS
     end
 
     # HtsFile
-    class SamHdr < FFI::Struct
+    class SamHdr < FFI::ManagedStruct
       layout \
         :n_targets,      :int32,
         :ignore_sam_err, :int32,
@@ -255,7 +255,7 @@ module HTS
 
     SamFile = HtsFile
 
-    class HtsTpool < FFI::Struct
+    class HtsTpool < FFI::ManagedStruct
       layout \
         :pool,           :pointer,
         :qsize,          :int
@@ -277,7 +277,7 @@ module HTS
         :next,           HtsOpt.ptr
     end
 
-    class HtsItr < FFI::BitStruct
+    class HtsItr < FFI::BitStruct # FIXME: ManagedBitStruct
       layout \
         :_flags,         :uint32, # bit_fields
         :tid,            :int,
@@ -358,14 +358,14 @@ module HTS
     class BamPileupCd < FFI::Union
       layout \
         :p,               :pointer,
-        :i,               :int64_t,
+        :i,               :int64,
         :f,               :double
     end
 
     class BamPileup1 < FFI::BitStruct
       layout \
         :b,              Bam1.ptr,
-        :qpos,           :int32_t,
+        :qpos,           :int32,
         :indel,          :int,
         :level,          :int,
         :_flags,         :uint32, # bit_fields
@@ -391,9 +391,9 @@ module HTS
         :line_skip,      :int32
     end
 
-    class Tbx < FFI::Struct
+    class Tbx < FFI::ManagedStruct
       layout \
-        :conf,           TbxConf.ptr,
+        :conf,           TbxConf,
         :idx,            HtsIdx.ptr,
         :dict,           :pointer
 
@@ -406,8 +406,8 @@ module HTS
 
     FaiFormatOptions = enum(:FAI_NONE, :FAI_FASTA, :FAI_FASTQ)
 
-    class Faidx < FFI::Struct
-      layout :bgzf,      BGZF,
+    class Faidx < FFI::Struct # FIXME: ManagedStruct
+      layout :bgzf,      BGZF.ptr,
              :n,         :int,
              :m,         :int,
              :name,      :pointer,
@@ -428,7 +428,7 @@ module HTS
     end
 
     # Complete textual representation of a header line
-    class BcfHrec < FFI::Struct
+    class BcfHrec < FFI::ManagedStruct
       layout \
         :type,           :int,
         :key,            :string,
@@ -463,7 +463,7 @@ module HTS
 
     class BcfIdinfo < FFI::Struct
       layout \
-        :info,           [:uint64_t, 3],
+        :info,           [:uint64, 3],
         :hrec,           [BcfHrec.ptr, 3],
         :id,             :int
     end
@@ -474,7 +474,7 @@ module HTS
         :val,            BcfIdinfo.ptr
     end
 
-    class BcfHdr < FFI::Struct
+    class BcfHdr < FFI::ManagedStruct
       layout \
         :n,              [:int, 3],
         :id,             [:pointer, 3], # BcfIdpair.ptr
@@ -536,7 +536,7 @@ module HTS
       layout \
         :pos,            :hts_pos_t,
         :rlen,           :hts_pos_t,
-        :rid,            :int32_t,
+        :rid,            :int32,
         :qual,           :float,
         :_n_info_allele, :uint32,
         :_n_fmt_sample,  :uint32,

data/lib/hts/libhts/cram.rb

@@ -352,7 +352,7 @@ module HTS
     # As int32_decoded/encode, but from/to blocks instead of cram_fd
     attach_function \
       :int32_put_blk,
-      %i[cram_block int32_t],
+      %i[cram_block int32],
       :int
 
     # Returns the refs_t structure used by a cram file handle.

data/lib/hts/libhts/sam.rb

@@ -242,13 +242,13 @@ module HTS
       [Bam1,
        :size_t,
        :string,
-       :uint16_t,
-       :int32_t,
+       :uint16,
+       :int32,
        :hts_pos_t,
-       :uint8_t,
+       :uint8,
        :size_t,
        :string,
-       :int32_t,
+       :int32,
        :hts_pos_t,
        :hts_pos_t,
        :size_t,

data/lib/hts/libhts/tbx.rb

@@ -68,3 +68,5 @@ module HTS
       :void
   end
 end
+
+require_relative "tbx_funcs"

data/lib/hts/libhts/tbx_funcs.rb

@@ -8,7 +8,7 @@ module HTS
       end
 
       def tbx_itr_queryi(tbx, tid, beg, end_)
-        hts_itr_query(tbx[:idx], tid, beg, end_, tbx_readrec)
+        hts_itr_query(tbx[:idx], tid, beg, end_, @@tbx_readrec)
       end
 
       def tbx_itr_querys(tbx, s)

data/lib/hts/libhts/vcf.rb

@@ -469,7 +469,7 @@ module HTS
     attach_function \
       :bcf_fmt_sized_array,
       [KString, :pointer],
-      :uint8_t
+      :uint8
 
     # Encode a variable-length char array in BCF format
     attach_function \